Tomer Cohen

He emphasises employee career growth aligns with the org’s need for agility, making change mutually beneficial.

Frame the shift as ongoing progress with no fixed end state so teams focus on routinely getting better than competitors.

Letting agents read the whole knowledge base failed, so they hand-selected high-quality examples, echoing LinkedIn feed lessons, to boost AI accuracy.

LinkedIn now assembles small cross-trained pods, akin to Navy SEALs, focused on a mission for a quarter before reassembling.

Hiring, calibration and biannual reviews now assess people on their AI agency and fluency to push early, active use of new tools.

Early results show top performers saving several hours weekly through the new tools, reinforcing their value.

Initially kept work to a small team; realised broader visibility and early tool access would have accelerated adoption.

Teams regularly demo useful AI tools in Slack and all-hands, encouraging exploration and peer learning.

Tomer says building clear cause-and-effect explanations unlocks iterative breakthroughs, echoing themes from The Beginning of Infinity.

LinkedIn launches Full Stack Builder pilots via small pods that must use tools and provide feedback to improve them.

Tomer argues great builders feel product pain deeply, caring like a parent and using healthy discontent to drive ongoing improvements.

Tomer cites Mindset's line 'becoming is better than being' to keep teams focused on continual growth and iteration rather than fixed states.

Granting AI access to all internal documents caused hallucinations, teaching the team to curate and converge on fewer tools.

Heads of each craft, like the head of Trust, construct domain-specific agents to embed their expertise into LinkedIn’s workflows.

Vision, empathy, communication, creativity and judgement stay with human builders while agents aim to automate every other task.

They build single-job expert agents first so they can rate and improve them, then plan to add an orchestrator layer later.